# Give staff agents one governed entry to many tools

> Give staff agents one governed entry to many tools: 0 companies fully do it, 6 with a limit, 127 carry the tag. Every listing in Agent tool access with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/aggregator.governed-gateway/
- Markdown: https://www.anchorterminal.com/niches/aggregator.governed-gateway/index.md (~1,200 tokens)
- Slim: https://www.anchorterminal.com/niches/aggregator.governed-gateway/index.min.md (~280 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/aggregator.governed-gateway/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 6 | 127 | 11 | 4 |

## What counts as done

One MCP endpoint that lists only allowed tools per agent or person, blocks or asks before writes, scans tool descriptions and arguments, and logs each call.

How agents ask for it: give the team's agents GitHub and Datadog; only allow read tools; block secrets leaving in tool calls

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Composio (API + MCP) (Sampark Inc (d/b/a Composio))**: Session config enables or disables toolkits and tools and sets require_approval by tool or by destructiveHint tag, and sessions are served over MCP. Description scanning wasn't found in the pages read, and acc-02 built its own gateway for finer permissions.
- **MCP Gateway (No legal entity found. COMMERCIAL.md, CLA.md and AUTHORSHIP.md name Mikko Parkkola, Finland, as licensor and author)**: One endpoint with per-key tool allowlists, a tool-poisoning validator on every description, an injection and credential-redaction firewall, rug-pull pinning and budgets. No approval step before writes and no per-call log export were found.
- **Merge Agent Handler (Merge API, Inc.)**: Tool access resolves per default, group and user, the Security Gateway scans every call's arguments to allow, redact or block, and logs export by API. No approval step before writes, and no scan of tool descriptions, was found.
- **One (One Systems, Inc.)**: Per-connection levels (read only, read and write, custom actions) are enforced on every call, execute_one_action is annotated destructive, and writes can ask first by MCP elicitation. No scan of tool descriptions or arguments was found.
- **StackOne (StackOne Technologies Limited)**: One MCP endpoint with Tool Policies that allow, mask fields or deny writes before the provider is called, Defender scanning responses for injection, PII redaction and request logs. Tool Policies are plan-gated and Defender scans responses, not descriptions.
- **Zapier MCP (agent actions) (Zapier Inc.)**: Separate read and write execute tools, account-level app and action restrictions, workspace task quotas and audit logs. Restrictions can't be set for MCP alone, and no description or argument scanning is documented.

- not this job: AgentWeb (Agent Web Labs, Inc.): AgentWeb runs its own site maps; it isn't a gateway in front of a team's own MCP servers or tools.
- not this job: letme (Long Lunch Labs Limited (company number 17523112, registered in England and Wales, incorporated 15 October 2026), trading as Anchor Terminal and letme): letme is not a gateway in front of a team's tools; it lists no tools per person and logs no tool calls.
- not this job: Monid (Monid, Inc.): Monid brokers third-party data endpoints; it doesn't front a team's own tools or list tools per person.
- not this job: Smithery (Clavia, Inc. (doing business as Smithery)): The proxy is gone, and acc-11 found 22 of 100 registry servers with unsafe tool descriptions when it ran.
- not this job: Unified.to MCP Server (Unified API Inc.): One MCP URL per customer connection with tool filters and hide_sensitive for PII; it is not a single governed entry for staff agents, and no write approval or description scanning is documented.

1 listings in the category were added after this reading and aren't marked yet.

## What stops agents finishing

- no per-call outbound authorisation from a sandbox
- registry servers with unsafe descriptions

## Around it

After: agent-observability.trace.

## Where this comes from

Weight 11 of 100 in Agent tool access: 4 of 32 accounts (HN gateway users and a registry scan)

As data: https://www.anchorterminal.com/api/v1/use-cases/aggregator.governed-gateway.json, or the MCP tool get_use_case with id aggregator.governed-gateway. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
